# Efficient description of experimental effects in amplitude analyses

## TL;DR

This paper reviews and proposes advanced density estimation techniques, including Gaussian processes and neural networks, to improve the treatment of instrumental effects in amplitude analyses of hadron decays.

## Contribution

It introduces novel applications of machine learning models for background estimation and efficiency correction in multidimensional amplitude analyses.

## Key findings

- Gaussian processes and neural networks effectively estimate densities in complex phase spaces
- Proposed methods improve background modeling accuracy in amplitude analyses
- Neural network-based density estimation offers flexible, model-assisted solutions

## Abstract

Amplitude analysis is a powerful technique to study hadron decays. A significant complication in these analyses is the treatment of instrumental effects, such as background and selection efficiency variations, in the multidimensional kinematic phase space. This paper reviews conventional methods to estimate efficiency and background distributions and outlines the methods of density estimation using Gaussian processes and artificial neural networks. Such techniques see widespread use elsewhere, but have not gained popularity in use for amplitude analyses. Finally, novel applications of these models are proposed, to estimate background density in the signal region from the sidebands in multiple dimensions, and a more general method for model-assisted density estimation using artificial neural networks.